Bristow and his colleague Stephen Liggett have since identified one gene variant that improves patients' response to bucindolol and another that is linked with heightened risk of side effects, according to Bristow, chief science and medical officer. The company, ARCA Discovery in Denver, will file for Food and Drug Administration approval within the next month or two and hopes to bring the drug to market in the summer of 2009.

Success stories remain rare, however, and coming up with personalized treatments is challenging, genomic scientists say. Researchers not only must identify genetic variations that are related to how a disease works, but figure out how they interact with a medication, as well.

Even when genomic knowledge can't save a drug, it can shed light on the reasons for failure and spur a new way of thinking about how a medication works, according to Scott Reines, a vice president in the pharmaceutical research-and-development unit of Johnson & Johnson. Companies could also develop and market tests that identify patients with a particular genetic variant.

"I think the bigger value isn't in going back and rescuing a particular drug, but you may be able to go back and rescue a particular line of thinking around a particular pathway," says Rosenkrans of the Personalized Medicine Coalition.